Does PPARγ 2 Gene Pro12Ala Polymorphism Affect Nonalcoholic Fatty Liver Disease Risk? Evidence from a Meta-Analysis

Abstract

Genetic factors can substantially contribute to the pathogenesis of nonalcoholic fatty liver disease (NAFLD). A missense Pro12Ala substitution in the PPARγ₂ gene (rs1801282) has been studied in relation with NAFLD risk in different ethnic groups, but findings have been inconclusive. The aim of this was to evaluate the association between rs1801282 and NAFLD through meta-analysis of all relevant published evidence. A systematic search to find eligible studies was performed in Medline, HuGE Navigator, and SCOPUS databases. The strength of association was evaluated using odds ratios with 95% confidence intervals obtained from a random effect approach and under additive, dominant, co-dominant, recessive, and allelic contrast models. Seven studies comprising 1474 cases and 2259 controls met the eligibility criteria and included in the meta-analysis. Combined results did not indicate any predisposing or protective effect for rs1801282 under any of the assessed modes of inheritance. The rate of heterogeneity was generally high due to the inter-study variations in terms of age, gender, and ethnicity. Evidence from the current meta-analysis indicated that rs1801282 variants are not associated with NAFLD risk. Future large-scale studies are required to substantiate the present findings.

Introduction

N onalcoholic fatty liver disease (NAFLD) is the most prevalent chronic hepatic disorder that encompasses a spectrum of pathological changes (Mohanty et al., 2009). Such a spectrum ranges from simple hepatic steatosis or mild hepatocyte inflammation (usually classified as grades I and II) to the more severe nonalcoholic steatohepatitis (NASH, classified as grade III or IV). With a global prevalence of 3%–24% (Williams, 2006; Amarapurkar et al., 2007) that is predicted to rise up to 25%–35% in future (Clark et al., 2002; Ruhl and Everhart, 2004; Farrell and Larter, 2006; Qureshi and Abrams, 2007), NAFLD is also regarded as an important public health burden. In recent years, the so-called two-hit theory has been adopted as the best model to explain the etiopathogenesis of NAFLD. Briefly, this model considers insulin resistance, increased fatty acid flux, reduced fatty acid β-oxidation, and excessive triglyceride accumulation in hepatocytes as the first hit, leading to hepatic steatosis. In the second hit, steatosed hepatocytes undergo oxidative stress, inflammation, and extensive injury that jointly promote disease severity into steatohepatitis, hepatic fibrosis, and finally hepatocellular carcinoma (Day and James, 1998). NAFLD is in close nexus with other metabolic disorders, including type 2 diabetes, obesity, and dyslipidemia, and is generally considered as the hepatic manifestation of the metabolic syndrome. The core feature in the pathogenesis of all aforementioned disorders is insulin resistance (Tsochatzis et al., 2009).

As with many other metabolic disorders, NAFLD has a polygenic nature and genetic factors have been suggested to exert substantial contribution to both development and progression stages (Daly et al., 2011; Hernaez, 2012). Peroxisome proliferator-activated receptor (PPAR) genes are among the susceptibility loci that have been studied in relation with NAFLD. PPARs are a group of nuclear transcription factors with α, β/δ, and γ subtypes. PPARγ is an important member of this family with pivotal regulatory effects on the expression of several factors involved in lipid metabolism and insulin resistance. Thus, modulators of PPARγ activity have attracted considerable attention as promising candidates for the treatment of NAFLD and associated metabolic disorders (Kallwitz et al., 2008).

The PPARγ gene is located on chromosome 3p25 and spans over 100 kbp. Transcription of the PPARγ gene is controlled by different promoters as well as alternative splicing. These complex transcriptional regulations lead to the expression of four isoforms of the protein, namely, PPARγ₁ to PPARγ₄ (Fajas et al., 1997; Gurnell, 2005). The rs1801282 is a missense single-nucleotide polymorphism (SNP) in the PPARγ₂ gene that stems from a C→G nucleotide substitution at the gene level, resulting in a Pro-to-Ala conversion at the codon 12 position of the translated protein. This SNP has been reported to be associated with decreased transcription (due to reduced DNA-binding affinity) and activity of PPARγ₂ and improved insulin sensitivity (Deeb et al., 1998). However, the impact of this SNP on the susceptibility to NAFLD is still controversial due to the inconsistent findings (Dongiovanni et al., 2010; Gupta et al., 2010; Rey et al., 2010; Cao et al., 2012; Gawrieh et al., 2012; Yang et al., 2012; Bhatt et al., 2013). The present meta-analysis aimed to provide a more comprehensive estimate of the association between PPARγ₂ rs1801282 SNP and NAFLD risk through meta-analysis of all relevant studies published to date.

Materials and Methods

Search strategy

A search strategy was devised to identify published studies investigating the association between the rs1801282 and NAFLD. The search followed the guidelines of the 2009 preferred reporting items for systematic reviews and meta-analysis statement and was undertaken in two stages (Moher et al., 2009). Initially, the citations in SCOPUS, MEDLINE, and HuGE Navigator databases were surveyed using the terms “peroxisome proliferator-activated receptorγ” or “PPARγ” or “peroxisome proliferator-activated receptorgamma” or “PPARgamma” or “PPARG”; these terms were combined with “polymorphism” or “polymorphic” or “SNP” or “genetic” or “genotype” or “variant” or “mutation” or “allele,” and all of the above terms were combined with “non-alcoholic” or “fatty liver” or “hepatic steatosis” or “steatohepatitis” or “NAFLD” or “NASH.” The wild-card term ‘‘*’’ was used to increase the sensitivity of the search strategy. Searches were limited to articles published in the English language. No previously published meta-analysis on the association between PPARγ₂ rs1801282 SNP and NAFLD was identified by searching in the HuGE Navigator database. Retrieved articles were hand-searched for any additional related studies.

Study selection

Only data from the full-text articles, not meeting or conference abstracts, were included. Studies that were included in the current meta-analysis were published up to November 2012, and (1) had a case–control design, (2) investigated the association between PPARγ₂ rs1801282 SNP and NAFLD, and (3) provided sufficient information on the frequency of genotypes in both case and control groups. Studies that (1) did not report complete data on the rs1801282 genotype frequencies, (2) did not fit the diagnostic criteria for NAFLD, (3) were not performed on humans, and (4) were conducted on patients with liver diseases due to alcohol abuse or viral infection were excluded.

Data extraction

From the eligible studies, the following data were extracted using standard protocols: (1) first author's name; (2) year of publication; (3) origin of studied population; (4) number of participants in the case and control groups; (5) age, gender, and body mass index (BMI) of study participants; (6) circulating concentration of hepatic transaminases [aspartate aminotransferase (AST) and alanine aminotransferase (ALT)] and lipid profile parameters [comprising total cholesterol, low-density lipoprotein cholesterol (LDL-C), high-density lipoprotein cholesterol (HDL-C) and triglycerides]; (7) systolic and diastolic blood pressure; (8) fasting glucose concentration; and (9) homeostasis model assessment-estimated insulin resistance (HOMA-IR) index. In case of incomplete data on the genotype frequencies, the authors of the respective article were contacted.

Genotyping

Genotyping of rs1801282 in the included studies was performed by polymerase chain reaction (PCR)–restriction fragment length polymorphism (Dongiovanni et al., 2010; Gupta et al., 2010; Cao et al., 2012; Bhatt et al., 2013) technique or TaqMan probe-associated real-time PCR (Rey et al., 2010; Gawrieh et al., 2012; Yang et al., 2012).

Quality assessment

The quality of each included study was methodologically evaluated based on the scale of Thakkinstian et al. (2005), with modifications in order to be adapted for NAFLD. Briefly, the quality assessment tool employed six items on the source of cases and controls, type of specimen for genotyping, method of NAFLD diagnosis, compliance with the Hardy–Weinberg equilibrium (HWE) in the control group, and population size. The overall score ranges from 0 to 18, with higher scores indicating higher qualities. Based on the overall score, quality of studies could be classified as low (score <6), medium (score 6–12), or high (score >12).

Quantitative data synthesis

Meta-analysis was conducted using the Cochrane Program Review Manager version 5.1. Combined unadjusted odds ratios (ORs) for the association of rs1801282 with NAFLD were generated under additive (GG vs. CC), dominant (GG + GA vs. CC), co-dominant (GA vs. GG + CC), recessive (GG vs. CC + CG), and allelic contrast (C allele vs. G allele) inheritance patterns. Meta-analysis was performed using the Mantel–Haenszel weighting method and under a random effect model due to the inter-study variations regarding age and demographic factors. Heterogeneity analysis was performed using the Cochran Q test and I² index. Sensitivity analyses were performed using the one-study remove approach to assess the impact of each study on the combined effect as previously described (Morris et al., 2012). The significance of the pooled ORs were assessed using Z-values. Compliance of genotype distribution with the HWE was assessed in the control group of each study using the χ² test with a threshold of p<0.05 to detect any disequilibrium. Random effect meta-regression was conducted to explore the association between BMI (as a major source of heterogeneity and an important determinant of NAFLD) and calculated ORs under different patterns of inheritance. In all tests, a p-value of<0.05 was considered as statistically significant.

Results

Flow of included studies

From 363 publications that were identified by initial database searches, 7 studies fulfilled inclusion criteria for the systematic review and were selected for further assessment. To identify these 7 articles, the full texts of 11 articles were evaluated. Four articles were found to be irrelevant due to the lack of case–control design (Kotronen et al., 2009; Delahanty et al., 2012) or not evaluating the target (rs1801282) SNP (Hui et al., 2008; Zhou et al., 2010) (Fig. 1).

FIG. 1.

Flow diagram of the study selection process.

Characteristics of included studies

Collectively, a total of 3733 individuals were included across the 7 eligible studies, comprising 1474 cases and 2259 controls. Included studies were conducted in India (Gupta et al., 2010; Bhatt et al., 2013), China (Cao et al., 2012; Yang et al., 2012), Italy (Dongiovanni et al., 2010), Germany (Rey et al., 2010), and United States (Gawrieh et al., 2012). The independent studies varied in population size between 179 (Gawrieh et al., 2012) and 903 (Yang et al., 2012) subjects. Overall, case and control groups in the included studies were matched regarding age and gender. Most of the studies failed to report demographic data on the frequency of important comorbidities such as cardiovascular disease, type 2 diabetes, hypertension, and hyperlipidemia. Besides, the distribution of other covariates such as lipid profile parameters and hepatic transaminases were not uniformly expressed across all studies. Reported data suggested that case patients had higher circulating levels of total cholesterol, triglycerides, LDL-C, AST, and ALT, while reduced levels of HDL-C, as would be expected in NAFLD. Consistently, plasma glucose and HOMA-IR index were higher in NAFLD patients than in controls. None of the included studies provided gender-specific data for the frequency of genotypes. Genotype frequencies in the control groups of all included studies were compliant with the HWE, apart from significant deviations that were observed in the studies by Bhatt et al. (2013) (χ²=12.51; p=0.004) and Gupta et al. (2010) (χ²=5.17; p=0.020) (Rey et al., 2010). Demographic characteristics of included studies' recruited populations are summarized in Table 1. Based on the overall calculated quality scores (Table 2), four of included studies had high quality (Dongiovanni et al., 2010; Rey et al., 2010; Cao et al., 2012; Yang et al., 2012), while the remaining three studies were classified as medium quality (Gupta et al., 2011; Gawrieh et al., 2012; Bhatt et al., 2013). Minor allele frequencies reported in the populations of included studies conformed to those previously reported by studies in the corresponding populations [retrieved from the ALFRED database (Rajeevan et al., 2003)] (Table 3).

Table 1.

Characteristics of Studies Included in the Meta-Analysis

Study		Bhatt et al.	Cao et al.	Dongiovanni et al.	Gupta et al.	Yang et al.	Rey et al.	Gawrieh et al.
Year		2013	2012	2010	2010	2012	2010	2012
Location		India	China	Italy	India	Chinese	Germany	USA
Employed diagnostic method		Ultrasonography	Medical history + clinical symptoms + laboratory findings + ultrasonography + liver biopsy (on suspected samples)	Liver biopsy	Raised transaminases or ultrasonography or liver biopsy (on suspected samples)	Liver biopsy	Liver biopsy	Liver biopsy
Participants	Case	162	169	202	98	436	263	212
	Control	173	699	346	280	467	259	62
Age (years)	Case	38.2±7.0	56.00 (49.00–65.50)^a	47.4±11	36±7	61.1±8.3	50.48±15.25	44±10
	Control	37.1±6.9	56.00 (46.00–64.00)	47.7±12	32±8	59.3±8.0	NR	43±13
Male (%)	Case	NR	24.85	80	67.35	35.77	NR	31.7
	Control	NR	25.61	79	65.00	44.11	NR	29.1
Smoker (%)	Case	NR	NR	NR	NR	NR	NR	NR
	Control	NR	NR	NR	NR	NR	NR	NR
BMI (kg/m²)	Case	28.1±3.2	26.59±3.08	27.4±3.8	26.9±4.5	26.6±3.0	NR	47±11
	Control	26.8±3.2	21.84±2.86	25.1±2.6	23.6±3.5	24.2±3.1	NR	44±13
Glucose (mM)	Case	4.93±0.66	6.37±2.09	5.44±1.33	5.77±2.33	5.9±1.3	NR	NR
	Control	4.83±0.59	5.71±1.68	4.91±0.55	ns	5.6±1.2	NR	NR
HOMA-IR	Case	2.5±0.2	2.84 (2.03–4.03)^a	4.1±3.3	2.7 (0.51–3.6)^a	2.30 (1.46–3.70)^a	NR	NR
	Control	1.6±0.3	1.20 (0.80–1.83)^a	2.9±1.6	NR	1.62 (1.03–2.44)^a	NR	NR
LDL-C (mM)	Case	2.86±0.59	2.74±0.93	NR	NR	3.2±0.8	NR	NR
	Control	2.71±0.60	2.64±0.90	NR	NR	3.2±0.7	NR	NR
HDL-C (mM)	Case	1.01±0.16	1.20 (1.02–1.49)^a	1.17±0.31	0.98±0.16	1.2±0.3	NR	NR
	Control	1.04±0.18	1.60 (1.36–1.90)^a	1.46±0.34	NR	1.4±0.4	NR	NR
Total cholesterol (mM)	Case	4.89±0.80	5.69±1.16	5.46±1.11	NR	5.5±1.1	NR	NR
	Control	4.63±0.76	5.44±1.09	5.02±0.88	NR	5.4±1.3	NR	NR
Triglycerides (mM)	Case	1.94±0.88	2.45±1.90	1.79±1.03	1.63 (0.66–11.07)^a	2.3±1.7	NR	2.58
	Control	1.67±0.87	1.28±0.99	1.06±0.49	NR	1.4±1.2	NR	1.98
AST (U/L)	Case	35.7±19.2	31.72±14.78	NR	NR	29.7±5.1	NR	NR
	Control	33.6±11.3	27.34±10.61	NR	NR	22.3±5.1	NR	NR
ALT (U/L)	Case	38.7±21.0	36.94±31.83	65±46	65±5.6	35.3±16.8	NR	62^b
	Control	35.0±13.7	17.88±9.56	23±9	27±4.6	25.8±15.2	NR	53^b
Hypertension (%)	Case	NR	NR	NR	25	NR	NR	56^b
	Control	NR	NR	NR	NR	NR	NR	25^b
CVD (%)	Case	0	NR	NR	NR	NR	NR	NR
	Control	0	NR	NR	NR	NR	NR	NR
Diabetes (%)	Case	0	NR	NR	27	20.87	NR	35
	Control	0	NR	NR	NR	15.42	NR	33
Hypercholesterolemia (%)	Case	NR	NR	NR	NR	NR	NR	NR
	Control	NR	NR	NR	NR	NR	NR	NR
SBP (mmHg)	Case	125.0±11.8	135.90±21.61	NR	NR	133.4±17.1	NR	NR
	Control	119.3±11.0	125.49±20.39	NR	NR	126.5±16.3	NR	NR
DBP (mmHg)	Case	80.2±8.6	84.00 (78.00–92.00)^a	NR	NR	84.0±10.3	NR	NR
	Control	77.4±8.6	80.00 (70.00–85.00)^a	NR	NR	81.2±16.9	NR	NR

Values are expressed as mean±SD unless otherwise stated.

Expressed as median (interquartile range).

No SD reported.

NR, not reported; BMI, body mass index; HOMA-IR, homeostasis model assessment-estimated insulin resistance; LDL-C, low-density lipoprotein; HDL-C, high-density lipoprotein; AST, aspartate aminotransferase; ALT, alanine aminotransferase; CVD, cardiovascular disease; SBP, systolic blood pressure; DBP, diastolic blood pressure.

Table 2.

Quality Assessment of Included Studies

Criterion	Subscale	Score	Bhatt et al. (2013)	Cao et al. (2012)	Dongiovanni et al. (2010)	Gupta et al. (2010)	Yang et al. (2012)	Rey et al. (2010)	Gawrieh et al. (2012)
Source of cases	Selected from population	3		3			3
	Selected from hospital	2	2			2
	Selected from pathology archives, but without description	1						3	1
	Not described	0			0
Source of controls	Population-based	3		3			3
	Blood donors or volunteers	2			2			2
	Hospital-based (cancer-free patients)	1							1
	Not described	0	0			0
Specimens used for determining genotypes	White blood cells or normal tissues	3	3		3	3	3
	Liver tissue or not described	0		0				0	0
Hardy–Weinberg equilibrium in controls	Hardy–Weinberg equilibrium	3		3	3		3	3	3
	Hardy–Weinberg disequilibrium	0	0			0
Diagnosis of NAFLD	Liver biopsy	3			3	1	3	3	3
	Ultrasonography	2	2	2
	Other	1
	Not described	0
Population size	>1000	3
	500–1000	2		2	2		2	2
	<500	1	1			1			1
	<200	0
Total score			8	13	13	7	17	13	9

NAFLD, nonalcoholic fatty liver disease.

Table 3.

Allele Frequencies of the rs1801282 Single-Nucleotide Polymorphism in Different Populations

rs1801282 MAF reported in the populations of included studies
Dongiovanni et al. (2010)		Rey et al. (2010)		Gupta et al. (2010)		Bhatt et al. (2013)		Yang et al. (2012)		Cao et al. (2012)		Gawrieh et al. (2012)
Italy		Germany		India		India		China		China		USA
NAFLD	Control	NAFLD	Control	NAFLD	Control	NAFLD	Control	NAFLD	Control	NAFLD	Control	NAFLD	Control
0.097	0.075	0.125	0.122	0.184	0.120	0.034	0.062	0.125	0.286	0.030	0.031	0.148	0.101

rs1801282 MAF reported in the populations of previously published studies
Geographic origin	Population	Typed sample size(2N) (Entry date in ALFRED)	ALFRED access ID	MAF
Region	Bariba	96 (12/11/2009)	SA003030G	0
Africa	Berba	98 (12/11/2009)	SA003031H	0
Africa	Amhara	174 (9/2/2009)	SA003001E	0.029
Africa	Oromo	158 (12/11/2009)	SA003029O	0.044
Africa	Croatian	1686 (4/15/2011)	SA003668X	0.160
Europe	French	1716 (10/20/2004)	SA000848U	0.105
Europe	Italians	408 (9/2/2009)	SA003004H	0.086
Europe	Italians	422 (9/2/2009)	SA003005I	0.092
Europe	Italians	782 (9/2/2009)	SA003006J	0.054
Europe	Italians	304 (9/2/2009)	SA003007K	0.046
Europe	Spaniards	790 (10/20/2004)	SA000852P	0.085
Europe	Pakistani	1584 (3/28/2011)	SA003636S	0.130
Asia	Tajik	200 (8/31/2009)	SA002952S	0.205
Asia	Sikh	772 (4/15/2011)	SA003667W	0.130
Asia	Han	414 (10/20/2004)	SA000851O	0.027
EastAsia	Han	296 (8/31/2009)	SA002953T	0.041
EastAsia	Koreans	506 (8/15/2006)	SA001822N	0.053
EastAsia	Cayapa	240 (9/2/2009)	SA003003G	0.004
SouthAmerica	Ecuadorian	106 (9/2/2009)	SA003002F	0

All data presented in this table have been retrieved from the ALelle FREquency Database (ALFRED) at http://alfred.med.yale.edu (accessed on January 19, 2013).

MAF, minor allele frequency.

Quantitative data synthesis

PPARγ₂ rs1801282 genotypes were investigated for possible association with NAFLD risk. Combined analysis was carried out using a random effect approach and under additive, dominant, co-dominant, recessive, and allelic contrast models. rs1801282 was not found to be associated with the risk of NAFLD under any of the assessed models (Fig. 2). Leave-one-out sensitivity analysis showed a significant and clinically relevant predisposing effect for the rs1801282 under additive [OR: 2.44; 95% confidence interval (CI): 1.18–5.02; p=0.02] and recessive [OR: 2.37; 95% CI: 1.16–4.88; p=0.02] models following omission of the study by Gawrieh et al. (2012) (Table 4).

FIG. 2.

Forest plot detailing unadjusted odds ratios (ORs) and 95% confidence intervals (CI) for the association between rs1801282 polymorphism and nonalcoholic fatty liver disease. The vertical line represents overall OR calculated in the current meta-analysis using a random effects model. (A–E) Findings of quantitative data synthesis under additive (GG vs. CC), dominant (GG + GA vs. CC), co-dominant (GA vs. GG + CC), recessive (GG vs. CC + CG), and allelic contrast (C allele vs. G allele) inheritance patterns, respectively. Meta-analysis was performed using the Mantel–Haenszel weighting method and under a random effect model. Heterogeneity analysis was performed using the Cochran Q test and I² index. The significance of the pooled ORs was assessed using Z-values.

Table 4.

Leave-One-Out Sensitivity Analysis of Combined Effect Size

Model	Statistics	Dongiovanni et al. (2010)	Rey et al. (2010)	Gupta et al. (2010)	Yang et al. (2012)	Gawrieh et al. (2012)	Bhatt et al. (2013)	Cao et al. (2012)
Additive	OR	1.50	1.73	1.54	1.79	2.44	1.82	1.79
	95% CI	0.50–4.49	0.41–7.28	0.56–4.27	0.68–4.72	1.18–5.02	0.41–8.11	0.68–4.72
	p-value	0.46	0.45	0.40	0.24	0.02	0.43	0.24
Dominant	OR	0.96	0.79	0.81	0.99	1.09	0.66	1.15
	95% CI	0.36–2.54	0.30–2.13	0.30–2.14	0.30–2.56	0.45–2.61	0.33–1.31	0.50–2.65
	p-value	0.93	0.65	0.66	0.99	0.85	0.23	0.73
Co-dominant	OR	0.84	0.89	0.79	0.98	1.00	0.82	0.87
	95% CI	0.53–1.33	0.55–1.44	0.54–1.16	0.66–1.46	0.70–1.43	0.53–1.27	0.56–1.36
	p-value	0.45	0.63	0.23	0.92	0.99	0.38	0.54
Recessive	OR	1.62	1.75	1.67	1.84	2.37	1.91	184
	95% CI	0.67–3.88	0.54–5.69	0.73–3.85	0.83–4.07	1.16–4.88	0.55–6.69	0.83–4.07
	p-value	0.28	0.36	0.23	0.13	0.02	0.31	0.13
Allelic contrast	OR	0.91	0.94	0.87	1.07	1.11	0.88	0.96
	95% CI	0.58–1.43	0.58–1.53	0.58–1.33	0.73–1.57	0.79–1.55	0.57–1.36	0.62–1.48
	p-value	0.67	0.81	0.52	0.73	0.54	0.58	0.85

Analysis was performed using a random effects model. OR, odds ratio; 95% CI, 95% confidence intervals. Significant results are bolded.

The rate of heterogeneity was generally high under analysis models (Table 5). Age, gender, ethnic, HOMA-IR, and BMI differences among the included studies are most plausible determinants of this high rate of heterogeneity.

Table 5.

Heterogeneity Analysis of Combined Effect Size

		Genotype			Allele
Model		CC	CG	GG	C	G
Additive	Tau²	—	—	0	—	—
	χ²	—	—	6.84	—	—
	df (Q)	—	—	6	—	—
	p	—	—	0.004	—	—
	I²	—	—	69%	—	—
Dominant	Tau²	—	1.17		—	—
	χ²	—	106.02		—	—
	df (Q)	—	6		—	—
	p	—	<0.00001		—	—
	I²	—	94%		—	—
Recessive	Tau²	—		0.16	—	—
	χ²	—		4.98	—	—
	df (Q)	—		4	—	—
	p	—		0.29	—	—
	I²	—		20%	—	—
Co-dominant	Tau²	—	0.20	—	—	—
	χ²	—	21.70	—	—	—
	df (Q)	—	6	—	—	—
	p	—	0.001	—	—	—
	I²	—	72%	—	—	—
Allelic contrast	Tau²	—	—	—	—	0.21
	χ²	—	—	—	—	27.57
	df (Q)	—	—	—	—	6
	p	—	—	—	—	<0.0001
	I²	—	—	—	—	78%

Q, Cochran Q test; df, degrees of freedom; I², I² index.

Since the included subjects in the study by Gawrieh et al. (2012) were morbidly obese, BMI values were further investigated as a source of heterogeneity. Random effect meta-regression between LogOR and BMI did not reveal any significant association under any of the evaluated models of inheritance, namely, additive (regression coefficient: −0.15; 95% CI: −0.202 to −0.090; τ²: 0.19; p=0.389), dominant (regression coefficient: −0.04; 95% CI: −0.181 to 0.106; τ²: 2.041; p=0.43), co-dominant (regression coefficient: −0.06; 95% CI: −0.010 to −0.011; τ²: 0.18; p=0.368), recessive (regression coefficient: −0.13; 95% CI: −0.179 to −0.073; τ²: 0.18; p=0.342), and allelic contrast (regression coefficient: −0.05; 95% CI: −0.097 to −0.002; τ²: 0.21; p=0.38).

Discussion

The present study represents the first meta-analysis on the association between PPARγ₂ rs1801282 SNP and risk of developing NAFLD. The results failed to indicate any protective or predisposing effect for the assessed SNP under any model of inheritance. Findings from the sensitivity analyses revealed that pooled estimates of OR were sensitive to the study of Gawrieh et al. (2012) under additive and recessive models. In the above-mentioned models, elimination of frequencies extracted from the referred study led to the estimation of a significant positive association between rs1801282 and NAFLD risk. This finding might imply a more complex pathology than NAFLD in the individuals recruited in the Gawrieh et al. (2012) study. As shown in Table 1, both case and control groups of the mentioned study had overwhelmingly high values of BMI in the morbid range. Although further metabolic and clinical chemistry data were not reported, such high BMI values represent an unfavorable metabolic status probably complicated by several morbidities even in the control group.

Given the pivotal role of PPARγ in the regulation of adipocyte differentiation, lipid accumulation and metabolism, and insulin sensitivity, its polymorphisms have been investigated in relation with a variety of metabolic traits. rs1801282 is a missense substitution of Ala for Pro, which results in decreased affinity to peroxisome proliferator response element and a modest impairment of transcriptional activation (Deeb et al., 1998; Masugi et al., 2000). It is widely accepted that PPARγ activation is a useful strategy to improve the global metabolic profile in high-risk patients. There is evidence linking PPARγ² activation with a multitude of favorable effects, including improvement of insulin sensitivity, enhancement of hepatic fatty acid β-oxidation, upregulation of adipokines including adiponectin, and mitigation of free fatty acid flux to the liver and hepatic inflammation (Kallwitz et al., 2008), though findings on PPAR activators such as thiazolidinediones have been equivocal due to the reported treatment-related hepatotoxicity (Scheen, 2001). rs1801282 has been the most studied PPARγ SNP, but findings of individual studies have been strongly equivocal. Collectively, the Ala12 allele appears to be associated with increased BMI, obesity insulin resistance, and risk of type 2 diabetes (Masud and Ye, 2003; Ochoa et al., 2004; Tellechea et al., 2009; Passaro et al., 2011; Ho et al., 2012). However, such an association does not appear to cause adverse metabolic profile (Gonzalez Sanchez et al., 2002; Passaro et al., 2011). The same controversy also exists among the findings on the association of this SNP with metabolic syndrome (Dongxia et al., 2008; Bego et al., 2011; Passaro et al., 2011). As for the NAFLD, the number of conducted clinical studies has been fewer, and all data on the association between rs1801282 and NAFLD have been published very recently between 2010 and 2013. Three out of seven included studies reported significant allelic difference between NAFLD and controls. Bhatt et al. (2013) and Gupta et al. (2010) indicated a higher prevalence of Ala12 in NAFLD patients in Asian Indians, while Yang et al. found an opposite finding in a Chinese population (Yang et al., 2012).

Given the multifactorial nature of NAFLD and contribution of several risk factors to its pathogenesis, it is most plausible that the contrasting findings are due to the differences in ethnic background, age, gender, environmental, and behavioral differences across the populations studied. Besides, NAFLD, the same as other closely associated metabolic disorders, arises from a complex interplay among several susceptibility loci (Daly et al., 2011; Hernaez, 2012). This multigenic origin necessitates a broader investigation taking into account the interactions of other SNPs within the PPARγ₂ gene or other nearby genes.

Aside from the issues of heterogeneity and single-polymorphism analysis, there are other limitations that deserve attention, though some are inherent to the meta-analysis. First, the number of included studies and the population size of each individual study were limited. Second, the pooled ORs were calculated without correcting for environmental covariates, due to the lack of access to the individual data. Third, the chance of publication bias still exists. This problem is due to the inclusion of published studies and lack of availability of any possible unpublished data. Finally, case and control groups were unmatched for metabolic factors in most of the included studies. It might be recommended for future investigations to be undertaken on metabolically matched subjects in order to control for potential confounding effects of different metabolic factors.

In conclusion, the present meta-analysis of seven case–control studies did not provide any evidence on the impact of rs1801282 within the PPARγ₂ gene on the susceptibility to NAFLD. While this finding is more robust and suggestive compared to conflicting individual genetic association studies, it is not conclusive. Hence, future large-scale trials are yet to be conducted in order to verify the present findings in larger and more homogenous populations that are matched for different environmental and metabolic factors.

Footnotes

Author Disclosure Statement

No competing financial interests exist.

References

Amarapurkar

D.N.

, Hashimoto

, Lesmana

L.A.

, Sollano

J.D.

, Chen

P.J.

, Goh

K.L.

2007. How common is non-alcoholic fatty liver disease in the Asia-Pacific region and are there local differences? J Gastroenterol Hepatol, 22:788–793.

Bego

, Dujic

, Mlinar

, Semiz

, Malenica

, Prnjavorac

, Ostanek

, Marc

, Causevic

2011. Association of PPARG and LPIN1 gene polymorphisms with metabolic syndrome and type 2 diabetes. Med Glas (Zenica), 8:76–83.

Bhatt

S.P.

, Nigam

, Misra

, Guleria

, Luthra

, Pandey

R.M.

, Pasha

M.A.

2013. Association of peroxisome proliferator activated receptor-gamma gene with non-alcoholic fatty liver disease in Asian Indians residing in north India. Gene, 512:143–147.

Cao

C.Y.

, Li

Y.Y.

, Zhou

Y.J.

, Nie

Y.Q.

, Wan

Y.J.

2012. The C-681G polymorphism of the PPAR-gamma gene is associated with susceptibility to non-alcoholic fatty liver disease. Tohoku J Exp Med, 227:253–262.

Clark

J.M.

, Brancati

F.L.

, Diehl

A.M.

2002. Nonalcoholic fatty liver disease. Gastroenterology, 122:1649–1657.

Daly

A.K.

, Ballestri

, Carulli

, Loria

, Day

C.P.

2011. Genetic determinants of susceptibility and severity in nonalcoholic fatty liver disease. Expert Rev Gastroenterol Hepatol, 5:253–263.

Day

C.P.

, James

O.F.

1998. Steatohepatitis: a tale of two “hits”? Gastroenterology, 114:842–845.

Deeb

S.S.

, Fajas

, Nemoto

, Pihlajamaki

, Mykkanen

, Kuusisto

, Laakso

, Fujimoto

, Auwerx

1998. A Pro12Ala substitution in PPARgamma2 associated with decreased receptor activity, lower body mass index and improved insulin sensitivity. Nat Genet, 20:284–287.

Delahanty

L.M.

, Pan

, Jablonski

K.A.

, Watson

K.E.

, McCaffery

J.M.

, Shuldiner

, Kahn

S.E.

, Knowler

W.C.

, Florez

J.C.

, Franks

P.W.

2012. Genetic predictors of weight loss and weight regain after intensive lifestyle modification, metformin treatment, or standard care in the Diabetes Prevention Program. Diabetes Care, 35:363–366.

10.

Dongiovanni

, Rametta

, Fracanzani

A.L.

, Benedan

, Borroni

, Maggioni

, Fargion

, Valenti

2010. Lack of association between peroxisome proliferator-activated receptors alpha and gamma2 polymorphisms and progressive liver damage in patients with non-alcoholic fatty liver disease: a case control study. BMC Gastroenterol, 10:102.

11.

Dongxia

, Qi

, Lisong

, Jincheng

2008. Association of peroxisome proliferator-activated receptorgamma gene Pro12Ala and C161T polymorphisms with metabolic syndrome. Circ J, 72:551–557.

12.

Fajas

, Auboeuf

, Raspe

, Schoonjans

, Lefebvre

A.M.

, Saladin

, Najib

, Laville

, Fruchart

J.C.

, Deeb

, Vidal-Puig

, Flier

, Briggs

M.R.

, Staels

, Vidal

, Auwerx

1997. The organization, promoter analysis, and expression of the human PPARgamma gene. J Biol Chem, 272:18779–18789.

13.

Farrell

G.C.

, Larter

C.Z.

2006. Nonalcoholic fatty liver disease: from steatosis to cirrhosis. Hepatology, 43:S99–S112.

14.

Gawrieh

, Marion

M.C.

, Komorowski

, Wallace

, Charlton

, Kissebah

, Langefeld

C.D.

, Olivier

2012. Genetic variation in the peroxisome proliferator activated receptor-gamma gene is associated with histologically advanced NAFLD. Dig Dis Sci, 57:952–957.

15.

Gonzalez Sanchez

J.L.

, Serrano Rios

, Fernandez Perez

, Laakso

, Martinez Larrad

M.T.

2002. Effect of the Pro12Ala polymorphism of the peroxisome proliferator-activated receptor gamma-2 gene on adiposity, insulin sensitivity and lipid profile in the Spanish population. Eur J Endocrinol, 147:495–501.

16.

Gupta

A.C.

, Chaudhory

A.K.

, Sukriti , Pande

, Sakhuja

, Singh

, Basir

S.F.

, Sarin

S.K.

2010. Peroxisome proliferators-activated receptor gamma2 Pro12Ala variant is associated with body mass index in non-alcoholic fatty liver disease patients. Hepatol Int, 5:575–580.

17.

Gurnell

2005. Peroxisome proliferator-activated receptor gamma and the regulation of adipocyte function: lessons from human genetic studies. Best Pract Res Clin Endocrinol Metab, 19:501–523.

18.

Hernaez

2012. Genetic factors associated with the presence and progression of nonalcoholic fatty liver disease: a narrative review. Gastroenterol Hepatol, 35:32–41.

19.

J.S.

, Germer

, Tam

C.H.

, So

W.Y.

, Martin

, Ma

R.C.

, Chan

J.C.

, Ng

M.C.

2012. Association of the PPARG Pro12Ala polymorphism with type 2 diabetes and incident coronary heart disease in a Hong Kong Chinese population. Diabetes Res Clin Pract, 97:483–491.

20.

Hui

, Yu-Yuan

, Yu-Qiang

, Wei-Hong

, Yan-Lei

, Xiao-Bo

, Yong-Jian

2008. Effect of peroxisome proliferator-activated receptors-gamma and co-activator-1alpha genetic polymorphisms on plasma adiponectin levels and susceptibility of non-alcoholic fatty liver disease in Chinese people. Liver Int, 28:385–392.

21.

Kallwitz

E.R.

, McLachlan

, Cotler

S.J.

2008. Role of peroxisome proliferators-activated receptors in the pathogenesis and treatment of nonalcoholic fatty liver disease. World J Gastroenterol, 14:22–28.

22.

Kotronen

, Yki-Jarvinen

, Aminoff

, Bergholm

, Pietilainen

K.H.

, Westerbacka

, Talmud

P.J.

, Humphries

S.E.

, Hamsten

, Isomaa

, Groop

, Orho-Melander

, Ehrenborg

, Fisher

R.M.

2009. Genetic variation in the ADIPOR2 gene is associated with liver fat content and its surrogate markers in three independent cohorts. Eur J Endocrinol, 160:593–602.

23.

Masud

, Ye

2003. Effect of the peroxisome proliferator activated receptor-gamma gene Pro12Ala variant on body mass index: a meta-analysis. J Med Genet, 40:773–780.

24.

Masugi

, Tamori

, Mori

, Koike

, Kasuga

2000. Inhibitory effect of a proline-to-alanine substitution at codon 12 of peroxisome proliferator-activated receptor-gamma 2 on thiazolidinedione-induced adipogenesis. Biochem Biophys Res Commun, 268:178–182.

25.

Mohanty

S.R.

, Troy

T.N.

, Huo

, O'Brien

B.L.

, Jensen

D.M.

, Hart

2009. Influence of ethnicity on histological differences in non-alcoholic fatty liver disease. J Hepatol, 50:797–804.

26.

Moher

, Liberati

, Tetzlaff

, Altman

D.G.

2009. Preferred reporting items for systematic reviews and meta-analyses: the PRISMA statement. BMJ, 339:b2535.

27.

Morris

D.R.

, Moxon

J.V.

, Biros

, Krishna

S.M.

, Golledge

2012. Meta-analysis of the association between transforming growth factor-beta polymorphisms and complications of coronary heart disease. PLoS One, 7:e37878.

28.

Ochoa

M.C.

, Marti

, Azcona

, Chueca

, Oyarzabal

, Pelach

, Patino

, Moreno-Aliaga

M.J.

, Martinez-Gonzalez

M.A.

, Martinez

J.A.

2004. Gene-gene interaction between PPAR gamma 2 and ADR beta 3 increases obesity risk in children and adolescents. Int J Obes Relat Metab Disord, 28,Suppl 3:S37–S41.

29.

Passaro

, Dalla Nora

, Marcello

, Di Vece

, Morieri

M.L.

, Sanz

J.M.

, Bosi

, Fellin

, Zuliani

2011. PPARgamma Pro12Ala and ACE ID polymorphisms are associated with BMI and fat distribution, but not metabolic syndrome. Cardiovasc Diabetol, 10:112.

30.

Qureshi

, Abrams

G.A.

2007. Metabolic liver disease of obesity and role of adipose tissue in the pathogenesis of nonalcoholic fatty liver disease. World J Gastroenterol, 13:3540–3553.

31.

Rajeevan

, Osier

M.V.

, Cheung

K.H.

, Deng

, Druskin

, Heinzen

, Kidd

J.R.

, Stein

, Pakstis

A.J.

, Tosches

N.P.

, Yeh

C.C.

, Miller

P.L.

, Kidd

K.K.

2003. ALFRED–the ALlele FREquency Database–update. Nucleic Acids Res, 31:270–271.

32.

Rey

J.W.

, Noetel

, Hardt

, Canbay

, Alakus

, Zur Hausen

, Dienes

H.P.

, Drebber

, Odenthal

2010. Pro12Ala polymorphism of the peroxisome proliferator-activated receptor gamma2 in patients with fatty liver diseases. World J Gastroenterol, 16:5830–5837.

33.

Ruhl

C.E.

, Everhart

J.E.

2004. Epidemiology of nonalcoholic fatty liver. Clin Liver Dis, 8:501–519vii.

34.

Scheen

A.J.

2001. Thiazolidinediones and liver toxicity. Diabetes Metab, 27:305–313.

35.

Tellechea

M.L.

, Aranguren

, Perez

M.S.

, Cerrone

G.E.

, Frechtel

G.D.

, Taverna

M.J.

2009. Pro12Ala polymorphism of the peroxisome proliferatoractivated receptor-gamma gene is associated with metabolic syndrome and surrogate measures of insulin resistance in healthy men: interaction with smoking status. Circ J, 73:2118–2124.

36.

Thakkinstian

, McEvoy

, Minelli

, Gibson

, Hancox

, Duffy

, Thompson

, Hall

, Kaufman

, Leung

T.F.

, Helms

P.J.

, Hakonarson

, Halpi

, Navon

, Attia

2005. Systematic review and metaanalysis of the association between β2-adrenoceptor polymorphisms and asthma: a HuGE review. Am J Epidemiol, 162:201–211.

37.

Tsochatzis

E.A.

, Manolakopoulos

, Papatheodoridis

G.V.

, Archimandritis

A.J.

2009. Insulin resistance and metabolic syndrome in chronic liver diseases: old entities with new implications. Scand J Gastroenterol, 44:6–14.

38.

Williams

2006. Global challenges in liver disease. Hepatology, 44:521–526.

39.

Yang

, Wen

, Li

, Tao

, Ye

, He

, Zhang

, Huang

, Chen

, Ling

, Qu

, Hu

2012. PPARG gene Pro12Ala variant contributes to the development of non-alcoholic fatty liver in middle-aged and older Chinese population. Mol Cell Endocrinol, 348:255–259.

40.

Zhou

Y.J.

, Li

Y.Y.

, Nie

Y.Q.

, Yang

, Zhan

, Huang

, Shi

S.L.

, Lai

X.B.

, Huang

H.L.

2010. Influence of polygenetic polymorphisms on the susceptibility to non-alcoholic fatty liver disease of Chinese people. J Gastroenterol Hepatol, 25:772–777.